[
https://issues.apache.org/jira/browse/ZOOKEEPER-1824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13970586#comment-13970586
]
Abhiraj Butala commented on ZOOKEEPER-1824:
-------------------------------------------
Thank you for the feedback Alex. I completely agree that we shouldn't
unnecessarily add one more method of generating an ensemble. I shall
investigate more on MiniZooKeeperCluster in hbase and get back. As per the
hbase jira mentioned here, the issue is also if zk should expose a public class
for creating an ensemble (and potentially reuse it), so hbase doesn't have to
redo it.
Thanks,
Abhiraj
> We need a MiniZooKeeperCluster for unit test
> --------------------------------------------
>
> Key: ZOOKEEPER-1824
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-1824
> Project: ZooKeeper
> Issue Type: Test
> Components: tests
> Reporter: chendihao
>
> We are developing a timestamp server to offer a precise auto increasing
> timestamp for other processes. ZooKeeper is used to select the master and
> store the persistent max offered timestamp.
> Now I'm not sure if the zookeeper servers work well no matter how much I
> damage the cluster randomly. So we need unit tests but ZooKeeper doesn't
> provides a single-process cluster for us. Should we implement the similar
> code like what HBase did in MiniZooKeeperCluster.java ?
> ref https://issues.apache.org/jira/browse/HBASE-2218
--
This message was sent by Atlassian JIRA
(v6.2#6252)
